Musk, DOGE, and the week in AI
Running the public purse is not running a manufacturing company. Plus DeepMind robotics, Anthropic's $3.5B raise, China's 'embodied AI', and VC consolidation.
Before we dive into the latest twists in AI, we should probably address what the actual Dogecoin is going on in the Trump administration. I'm all for smarter government — but really?
Fun side-quest or careful Machiavellian execution? Either way, Musk has established himself as a champion of efficiency at the very fulcrum of Western economic power. His fixation on streamlined processes and minimal overhead — honed lowering the cost of EVs and space travel at Tesla and SpaceX — has been transposed into government, professing to save us from inflated public-administration costs. We'll temporarily suspend the fact that EVs have increased the cost of the basket of goods, and that self-serving billionaire asset-hoarding tends to push up global prices.
With a net worth approaching $400 billion, Musk controls stakes in companies valued at over $1 trillion collectively — Tesla's ~$620B cap, SpaceX's ~$140B valuation — letting him amass vast resources, from 6,000 acres for Tesla factories to a $100M investment in xAI's Memphis data centre.
The thing about running an empire, though, is it's not just about cutting costs. The most effective leaders foster a sense of purpose beyond the bottom line — efficiency isn't the endgame, it's the stage on which a more complex play unfolds. This might work at Tesla; my fear with government is that it misguides an entire electorate into thinking the public purse is as conceptually simple as a heavy-goods manufacturer. It's not, and never will be.
AI updates
xAI's acquisition of a million-square-foot Memphis data centre isn't just real estate; it's part of a strategy to deepen influence over the technologies shaping our future. This isn't about following trends, it's about creating them. Google DeepMind introduced two advanced models — Gemini Robotics and Gemini Robotics-ER — bringing more advanced reasoning to general-purpose robots, enabling complex actions like folding origami, arranging workspaces, even playing basketball. A step change in real-world adaptability.
Anthropic announced it raised $3.5 billion at a $61.5 billion post-money valuation, led by Lightspeed. Staggering sums. At China's National People's Congress, leaders spotlighted AI — Premier Li Qiang highlighting 'embodied AI' to drive domestic consumption — with a sharp increase in tech funding in the 2025 budget. AI tools are also increasingly used to detect errors in research papers: the Black Spatula Project has analysed 500 papers and YesNoError over 37,000, though experts stress the need for human verification.
The shifting landscape of AI economics
US power-sector M&A is surging, driven by data-centre energy demand — Constellation Energy's $16.4 billion acquisition of Calpine being a marker. Alibaba announced a massive 380 billion yuan ($52.4B) investment in AI and cloud over three years, the largest private computing-infrastructure investment ever seen in China. Nvidia, which lost over $1 trillion in market value early in 2025 amid DeepSeek fears and a viral short thesis, may be eyeing a bounce-back as long-term infrastructure demand holds.
VC consolidation
The trend toward consolidation continues: nine US AI startups have each raised over $100M in 2025, with just six — OpenAI, Databricks and xAI among them — taking a staggering 40% of total Q4 funding. Will that concentration drive more breakthroughs, or stifle competition? Meanwhile, industry veteran Larry Page is back with a new venture, Dynatomics, applying AI to product manufacturing — a reminder there's still room for fresh ideas in an ever-expanding landscape.